SYSTEMIC RISK Jean TIROLE The Future of Financial Regulation Banque de France - TSE conference January 28, 2009
SYSTEMIC RISK
Jean TIROLE
The Future of Financial Regulation
Banque de France - TSE conference January 28, 2009
I. INTRODUCTION
(At least) two meanings of systemic risk:
(a) Domino e¤ectsCreated by:
� mutual exposures[OTC markets, money market, etc.]
� absence of centralized exchange[Amaranth 2006 vs. Lehman or AIG 2008].
Highly endogenous.Peer monitoring and domino e¤ects two sides of the same coin?
2
(b) Overall macroeconomic fragility
X Increased reliance on markets (securitization, money market):
� traditional institutions (commercial banks, broker-dealers)� shift from bank-based system to market-based system (shadow bankingsystem: conduits, hedge funds, investment banks, monolines;mutual funds: threats of redemptions).
X Subprime borrowers, LBOs.
X Strong monetary policy reaction:
� $4,400 billion commitments through various Fed facilities� Fed funds rate almost 0.
3
(b) Overall macroeconomic fragility
X Increased reliance on markets (securitization, money market):
� traditional institutions (commercial banks, broker-dealers)� shift from bank-based system to market-based system (shadow bankingsystem: conduits, hedge funds, investment banks, monolines;mutual funds: threats of redemptions).
X Subprime borrowers, LBOs.
X Strong monetary policy reaction:
� $4,400 billion commitments through various Fed facilities� Fed funds rate almost 0.
4
Emmanuel Farhi-Jean Tirole"Leverage and the Central Banker�s Put" (AER P&P 2009)+ larger paper (in progress)
Transformation, interest rate vulnerability and monetary bailout
5
Key insight
Formal relationship between monetary developments andmarket-based �nancing:
"monetary policy" (interest rates)
� limited commitment� not targeted
the more economic actors exhibit interest-rate vulnerability, the morethe central bank must keep interest rates low.
Private leverage choices strategic complements through reactionof monetary policy
6
Key insight
Formal relationship between monetary developments andmarket-based �nancing:
"monetary policy" (interest rates)
� limited commitment� not targeted
the more economic actors exhibit interest-rate vulnerability, the morethe central bank must keep interest rates low.
Private leverage choices strategic complements through reactionof monetary policy
7
Key insight
Formal relationship between monetary developments andmarket-based �nancing:
"monetary policy" (interest rates)
� limited commitment� not targeted
the more economic actors exhibit interest-rate vulnerability, the morethe central bank must keep interest rates low.
Private leverage choices strategic complements through reactionof monetary policy
8
Corollaries
Private leverage decisions highly sensitive to macroeconomicconditions.
Private sector interest-rate vulnerability may increase when bad newsabout future liquidity needs accrue.
Monetary policy time-inconsistent, not for standard in�ation-biasreason.
Need for macro-prudential supervision.
9
II. DESCRIPTION OF MODEL
Choice of exposure to liquidity risk = choice between twotechnologies: "risky" and "safe". In practice:
reliance on securitization.
lack of hedging (purchase of protection),
failure to hoard assets or secure lines of credit.
Risky technology allows larger investment.
10
II. DESCRIPTION OF MODEL
Choice of exposure to liquidity risk = choice between twotechnologies: "risky" and "safe". In practice:
reliance on securitization.
lack of hedging (purchase of protection),
failure to hoard assets or secure lines of credit.
Risky technology allows larger investment.
11
II. DESCRIPTION OF MODEL
Choice of exposure to liquidity risk = choice between twotechnologies: "risky" and "safe". In practice:
reliance on securitization.
lack of hedging (purchase of protection),
failure to hoard assets or secure lines of credit.
Risky technology allows larger investment.
12
CENTRAL BANK
Objective function
Can distort consumers�marginal rate of substitution between dates 1and 2, and lower (one plus) rate of interest R < R�. Distortion cost(MRS 6=MRT): L( R
(�)) = loss
13
CENTRAL BANK
Objective function
Can distort consumers�marginal rate of substitution between dates 1and 2, and lower (one plus) rate of interest R < R�. Distortion cost(MRS 6=MRT): L( R
(�)) = loss
14
MONETARY BAILOUTS
If �rm
has no cash at date 1 (extreme case: no store of value between dates0 and 1)
faces liquidity need,
then it must issue new securities on date-2 cash �ow.
Re�nancing constraint:
pledgeable incomeR
� reinvestment need (*)
Lower interest rate facilitates re�nancing. Let R0 < R� satisfy (*)with equality.
15
MONETARY BAILOUTS
If �rm
has no cash at date 1 (extreme case: no store of value between dates0 and 1)
faces liquidity need,
then it must issue new securities on date-2 cash �ow.
Re�nancing constraint:
pledgeable incomeR
� reinvestment need (*)
Lower interest rate facilitates re�nancing. Let R0 < R� satisfy (*)with equality.
16
∆ � β [bene�t for �rm from being able to continue]
�[reinvestment need� pledgeable incomeR �
]
X Bene�t of low interest rate policy is proportional to number of �rmsthat have chosen risky technology and are in distress:
X Cost of low interest rate policy L(R0) is �xed (independent of x)
Non-targeted policy
17
∆ � β [bene�t for �rm from being able to continue]
�[reinvestment need� pledgeable incomeR �
]
X Bene�t of low interest rate policy is proportional to number of �rmsthat have chosen risky technology and are in distress:
X Cost of low interest rate policy L(R0) is �xed (independent of x)
Non-targeted policy
18
∆ � β [bene�t for �rm from being able to continue]
�[reinvestment need� pledgeable incomeR �
]
X Bene�t of low interest rate policy is proportional to number of �rmsthat have chosen risky technology and are in distress:
X Cost of low interest rate policy L(R0) is �xed (independent of x)
Non-targeted policy
19
Result #1. If ∆γx � L(R0)monetary bailout.
Strategic complementarities
Result #2. Time inconsistency(di¤erent from standard TI, à la Kydland-Prescott).
Result #3 Monetary bailout
Risky corporate behavior more likely if
higher weight of corporate sector in central bank�s objective (β high)
liquidity shocks more likely (γ high) ! contrast with commitmentsolution.
20
Result #1. If ∆γx � L(R0)monetary bailout.
Strategic complementarities
Result #2. Time inconsistency(di¤erent from standard TI, à la Kydland-Prescott).
Result #3 Monetary bailout
Risky corporate behavior more likely if
higher weight of corporate sector in central bank�s objective (β high)
liquidity shocks more likely (γ high) ! contrast with commitmentsolution.
21
Result #1. If ∆γx � L(R0)monetary bailout.
Strategic complementarities
Result #2. Time inconsistency(di¤erent from standard TI, à la Kydland-Prescott).
Result #3 Monetary bailout
Risky corporate behavior more likely if
higher weight of corporate sector in central bank�s objective (β high)
liquidity shocks more likely (γ high) ! contrast with commitmentsolution.
22
Regulating leverage
X With commitment, no need for leverage regulation:
� privately and socially sub-optimal to start a risky project that is dis-continued in case of distress
� privately and socially optimal to choose risky project if monetary bailout
X Without commitment, leverage regulation to alleviate temptation tobail out.
Macro-prudential regulation
23
Regulating leverage
X With commitment, no need for leverage regulation:
� privately and socially sub-optimal to start a risky project that is dis-continued in case of distress
� privately and socially optimal to choose risky project if monetary bailout
X Without commitment, leverage regulation to alleviate temptation tobail out.
Macro-prudential regulation
24
FISCAL BAILOUTS
X Di¤erent kind of ine¢ ciency. Can be targeted. But (due toinformational limitations) consumer money may end up subsidizing�rms that don�t need it.
X Monetary and �scal bailouts. May be complements.
25
FISCAL BAILOUTS
X Di¤erent kind of ine¢ ciency. Can be targeted. But (due toinformational limitations) consumer money may end up subsidizing�rms that don�t need it.
X Monetary and �scal bailouts. May be complements.
26